That's understandable. The main site for those ratings, eloratings.net, is up and down like your mom's underwear, so I'll explain here.
Imagine you have a standard team. Let's call them Norway for the sake of it. If, say, Ireland were to play Norway a hundred times, I'd die of boredom. Also, I'd expect Ireland to maybe score 52% (counting a draw as a half a win). If Spain played Norway 100 times, they might score about 85%. The Elo system reckons that that means Spain would beat Ireland a heck of a lot of the time too, and quantifies that.
Of course, no one plays any one team all that often, so the scheme basically updates the rating of each team after each match with the result: if Ireland loses to Spain, Ireland's rating goes down a little bit ans Spain's goes up a little. If the opposite result happens, bigger changes happen. If Ireland beats Norway, we go up by a moderate amount - smaller than for beating Spain, but bigger than Spain would get for the same result. So, over time, everyone's rating tends to wobble near their true strength.
The wiki link I gave has the current table for the top 60 teams in the world.
The only weakness in the whole thing is that it assumes a normal distribution to results by rating difference, which is only approximately true. It's a far better ranking than anything FIFA's ever come up with.